Summary
Frequent mutations of the metabolic enzymes isocitrate dehydrogenase 1 (IDH1) and IDH2 are found in multiple types of cancer. The mutations result in the accumulation of 2HG, a putative “oncometabolite” which is believed to mediate the oncogenic potential of mutant IDH. However, it is unclear whether mutant IDH is sufficient for tumorigenesis in vivo and the role of 2HG in the transformation process. Furthermore, despite strong evidence linking IDH mutation to DNA and histone hypermethylation, it remains to be shown that epigenetic misregulation contribute to the malignant phenotype caused by mutant IDH. Here we report that in a non-transformed murine mesenchymal multipotent cell line, expression of an IDH mutant impaired mesenchymal differentiation, induced loss of contact inhibition and generated highly mitotic mesenchymal tumors in xenograft study. DNA methylation profiling at base-resolution revealed a profound genome-wide CpG hypermethylation in IDH mutant cells which is enriched for genes involved in cell- cell contact. Our data suggest that epigenetic abnormality and the resultant aberrant gene expression represent a critical step in the initiation and progression of tumors with IDH mutation.
Introduction
Since the original observation by the German biochemist Otto Warburg that cancer cells utilize glucose in a fundamentally different way from normal cells (Warburg, 1956), the recent resurgence in cancer metabolism research led to the increasing appreciation that metabolic reprogramming is a hallmark of cancer (Vander Heiden et al., 2009; Ward and Thompson, 2012). However, it remains controversial whether metabolism rewiring observed in cancer plays a significant role in driving cancer development. A strong argument in support of this hypothesis is the recent identification of cancer-associated germline and somatic alterations of genes encoding for metabolic enzymes (Mullarky et al., 2011; Oermann et al., 2012), including prevalent mutations in isocitrate
dehydrogenase 1 (IDH1) and IDH2.
Cytosolic IDH1 and mitochondrial IDH2 are NADP+ -dependent enzymes that
metabolize isocitrate to αKG. Frequent somatic mutations of IDH1 and IDH2 were
identified in intermediate-grade gliomas (Yan et al., 2009), adult de novo acute myeloid leukemias (AMLs) (Mardis et al., 2009; Ward et al., 2010) and subsets of
chondrosarcomas, lymphomas and cholangiocarcinomas (Amary et al., 2011; Borger et al., 2012; Cairns et al., 2012). The fact that mutations observed in IDH1 and IDH2 are always heterozygous point mutations affecting only a few residues suggests that they are unlikely loss of function. Indeed, metabolomic and biochemical analysis revealed that mutant IDHs gain a neomorphic activity of producing 2-hydroxyglutarate (2HG) from
αKG (Dang et al., 2009; Ward et al., 2010). 2HG is normally present at very low levels in
mutation. It is believed that IDH mutations promote tumorigenesis through accumulating the putative “oncometabolite” 2HG.
At the molecular level, mounting evidence implicate a link between IDH mutation and epigenetic dysregulation, particularly DNA hypermethylation. In multiple malignancies, IDH mutations are tightly associated with a CpG island methylator phenotype (CIMP) in patient samples (Figueroa et al., 2010a; Noushmehr et al., 2010; Pansuriya et al., 2011; Wang et al., 2012), and expression of an IDH mutant in cell lines is sufficient to cause DNA hypermethylation (Figueroa et al., 2010a; Turcan et al., 2012). Mechanistically,
mutant IDH was shown to impair activities of αKG-dependent chromatin modifying
enzymes by producing 2HG as a competitive inhibitor. These include TET family
enzymes which converts 5-methylcytosine (5mC) to 5-hydroxymethylcytosine (5hmC), a novel epigenetic mark potentially involved in DNA demethylation (Figueroa et al., 2010a); as well as Jumonji-C histone demethylases (Chowdhury et al., 2011; Lu et al., 2012). Importantly, genetic profiling of AML samples revealed a mutual exclusivity between IDH mutations and TET2 loss of function mutations, providing strong genetic
evidence that IDH mutation may support tumorigenesis through modulating αKG-
dependent chromatin modifier and the resultant hypermethylation of DNA and histone.
Despite the advance in understanding the molecular mechanism, it is yet to be shown whether mutant IDH is sufficient for tumorigenesis. Here we report that in a non- transformed murine mesenchymal progenitor cell line, IDH mutation leads to impaired mesenchymal lineage differentiation and loss of contact inhibition. Importantly, mutant IDH cells form highly mitotic tumors in mouse xenograft study. DNA methylation sequencing at base-pair resolution showed that IDH mutant cells display genome-wide
DNA hypermethylation, including key genes involved in cell adhesion. Collectively our data suggest that increased DNA methylation represents a critical step in the
Results
IDH mutation inhibits mesenchymal lineage differentiation
We have previously reported that IDH mutation could block the adipocyte differentiation from non-transformed mouse 3T3-L1 fibroblasts (Lu et al., 2012). To determine the generality of these findings, we retrovirally transduced C3H10T1/2 (10T) cells with vectors containing either wild-type or R172K mutant IDH2. 10T cells were originally isolated from C3H mouse embryo. They were sensitive to confluence-induced
proliferation arrest and showed no tumorigenicity in xenograft study (Reznikoff et al., 1973). Subsequently, 10T cells were demonstrated to be multipotent with the ability to differentiate into several mesenchymal lineages, including adipocytes, myoblasts and chondrocytes (Taylor and Jones, 1979). Notably, IDH mutations are found at high frequencies in chondrocytic sarcomas (Amary et al., 2011). 10T cells expressing R172K mutant IDH2 but not empty vector or wild-type (WT) IDH2 showed significant increase in 2HG levels (Figures 4.1A and 4.1B).
Consistent with previous findings, upon adipogenesis induction, IDH2 mutant cells showed a profound impairment in adipocyte differentiation with a lack of lipid droplets accumulation and absence of adipocyte-specific gene expression (Adipoq and Fabp4) compared to vector and WT IDH2 cells (Figure 4.1C). Similarly, when cells were subjected to conditions that promote chondrogenesis, morphological conversion from fibroblast-like to rounded shapes resembling mature chondrocytes and expression of chondrocyte-specific markers Acan and Col2a1 were only observed in vector and WT IDH2 cells but not IDH2 mutant cells (Figure 4.1D). Furthermore, while vector and WT
Figure 4.1:IDH mutation inhibits mesenchymal lineage differentiation.
(A) 10T cells were retrovirally transduced with empty vector or vectors containing either WT or R172K mutant IDH2. After puromycin selection, levels of IDH2 protein were assessed by Western blot. (B) Levels of 2HG measured by GC-MS in 10T cells expressing vector, WT or mutant IDH2. (C) Vector, WT or mutant IDH2 cells were treated with adipogenesis cocktail. Microscopic images of cell morphology were recorded 7 days after differentiation. mRNA expression of Adipoq and Fabp4 was measured by Q-RTPCR. (D) Vector, WT or mutant IDH2 cells were treated with
chondrogenesis cocktail. Microscopic images of cell morphology were recorded 10 days after differentiation. mRNA expression of Acan and Col2a1 was measured by Q-RTPCR.
IDH2 cells became growth arrested after the induction of adipocyte or chondrocyte
differentiation, IDH2 mutant cells continued to proliferate (Figure4.2A) and maintained
high levels of cyclin D1 (Figure 4.2B). Taken together these data suggest that differentiation blockade by IDH mutation is a common feature shared by multiple mesenchymal lineages.
IDH mutant cells escape contact inhibition and generate tumors in vivo
The differentiation experiments described above require cells to be seeded at confluence before induction. Therefore the observations that IDH mutant cells were able to continue proliferation at post-confluence prompted us to determine whether these cells have become insensitive to contact inhibition. When cultured in normal growth medium, vector, WT IDH2 or IDH2 mutant cells showed no difference in proliferation rate at sparse (day 0 – day 2, Figure 4.3A). However, while vector and WT IDH2 cells stopped proliferation after reaching confluence (day 4 – day 6), the accumulation of IDH2 mutant cells continued even after confluency was achieved. Cell cycle analysis showed that IDH2 mutant cells had a higher percentage of cell population in G2 and S phase at post- confluence (Figure 4.3B). Moreover, the induction of cell cycle inhibitor p27 and the decrease in levels of cyclin D1 after contact inhibition were less pronounced in IDH2 mutant cells (Figure 4.3C).
IDH2 mutant cells underwent cell cycle arrest normally after serum deprivation (Figure 4.3D), suggesting that their escape from contact inhibition is unlikely due to defects in cell cycle regulation. The cadherins are critical components of cell-cell adhesion and have been proposed to be the main upstream mediator of contact inhibition
Figure 4.2:Differentiation blockade led to sustained proliferation of IDH mutant cells.
(A) 10T vector, WT or mutant IDH2 cells were treated with adipogenesis or
chondrogenesis cocktail. Cell numbers were counted at day 0, 3, 6 after differentiation induction. (B) 10T vector, WT or mutant IDH2 cells were treated with adipogenesis or chondrogenesis cocktail. Six days after differentiation, cells were lysed and protein levels of cyclin D1 and p27 were measured by Western blot.
Figure 4.3: IDH mutant cells escape contact inhibition.
(A) 10T vector, WT or mutant IDH2 cells were cultured in normal growth medium and cell numbers were counted at day 0, 2, 4 and 6. Cells reached confluence at day 4. (B) FACS cell cycle analysis of vector, WT or mutant IDH2 cells at post-confluence day 2. (C) Vector, WT or mutant IDH2 cells were lysed at sparse, at confluence or 2 days post- confluence. Protein levels of cyclin D1 and p27 were measured by Western blot. (D) FACS cell cycle analysis of vector, WT or mutant IDH2 cells at sparse in growth medium with 10% or 0.1% fetal calf serum. (E) Vector, WT or mutant IDH2 cells were lysed at sparse or 2 days post-confluence. Levels of N-cadherin protein expression were measured by Western blot and mRNA expression was measured by Q-RTPCR.
(Gumbiner, 2005). We found that unlike vector and WT IDH2 cells, IDH2 mutant cells failed to upregulate N-cadherin protein and mRNA expression after contact inhibition (Figures 4.3E), suggesting that the post-confluence growth of these cells might be the result of inability to sense cell-cell contact at surface membrane.
Since loss of contact inhibition is a classical feature of tumor cells and often indicates transformation, we next tested whether 10T cells with IDH mutation generate tumors in vivo. IDH2 mutant cells formed palpable tumors after 20 days of subcutaneous injection
and continued to grow up to 1000 m3, while vector and WT IDH2 cells showed no
tumorigenecity over the course of study (Figure 4.4A). Immunohistochemistry staining revealed that IDH2 mutant tumors were highly mitotic with high Ki67 index (Figure 4.4B). Consistent with cell culture study, these tumors also had high levels of cyclin D1 and low p27 expression.
IDH mutation induces DNA hypermethylation
We next sought to determine the mechanism of tumorigenesis by IDH mutation. Previous
reports implicate a role of 2HG in inhibiting αKG-dependent chromatin modifying
enzymes including TET2 and jumonji-C histone demethylases. These enzymes belong to
the superfamily of αKG-dependent dioxgenases, which consist of ~100 members
covering diverse biological functions. Among them, 2HG has been proposed to affect
activities of prolyl hydroxylases 2 (PHD2) which hydroxylases HIF1α for degradation
(Koivunen et al., 2012; Zhao et al., 2009). Moreover, in a knock-in IDH1 mutant mouse model, 2HG was shown to disrupt collagen processing and cause endoplasmic reticulum (ER) stress response by inhibiting collagen prolyl-hydroxylation (Sasaki et al., 2012a).
Figure 4.4: IDH mutant cells generate tumors in vivo.
(A) 10 millions of 10T vector, WT or mutant IDH2 cells at passage 20 were
subcutaneously injected into nude mice. Tumor implantation and growth were monitored and measured. Images are shown for mice injected with WT cells (left) or mutant cells (right). (B) IHC analysis was performed on mutant IDH2 tumors and representative images are shown for H&E, Ki67, cyclin D1 and p27.
We found that soluble fraction of collagen IV seemed to increase in IDH2 mutant cells with a faster migration pattern on SDS gel, suggesting a possible reduced hydroxylation and delayed processing of collagen IV (Figure 4.5A). In addition, we observed a
decreased expression of collagen I. However, coating cell culture plates with collagen did not blunt the post-confluence proliferation of mutant IDH2 cells (Figure 4.5B). There was no difference in the levels of ER stress response between IDH2 mutant and WT cells either (Figure 4.5C), suggesting that the mild defect in collagen secretion is unlikely to
account for the tumorigenecity of IDH2 mutant cells. Finally, the HIF1α accumulation
under hypoxia seemed to be unaffected by IDH2 mutation (Figure 4.5D).
When we extracted DNAs from cells cultured at sparse and subjected them to extended reduced representation bisulfite sequencing (ERRBS) to obtain base-resolution DNA methylation profiles (Akalin et al., 2012), IDH2 mutant cells showed a genome-wide distribution of differentially methylated CpG sites with a profound hypermethylation across all chromosomes (Figure 4.6A). Among the 2,400 genes with differentially
methylated cytosines at the promoters, we observed a predominance in hypermethylation which was highly statistical significant (78.1 % being hypermethylated, p-value < 0.0001 by Chi-square test). In contrast, significantly less differentially methylated cytosines were observed between WT IDH2 and vector cells with even distribution of hypermethylated and hypomethylated sites (Figure 4.6B). Pathway enrichment analysis of the
hypermethylated sites between IDH2 mutant and WT cells revealed genes involved in cell adhesion and WNT signaling pathways which are implicated in the response to cell- cell contact (Figure 4.6C). Finally, we also found increased methylation at several histone marks such as H3K9me3, H3K9me2 and H3K4me3 in IDH2 mutant cells, which could
Figure 4.5: Collagen processing and HIF signaling in IDH mutant cells.
(A) 10T vector, WT or mutant IDH2 cells were lysed and protein levels of collagen I and collagen IV in the solution fraction were measured by Western blot. (B) 10T vector, WT or mutant IDH2 cells were seeded on normal cell culture plates or plates coated with collagen and cell numbers were counted at day 0, 2, 4 and 6. Cells reached confluence at day 4. (C) Vector, WT or mutant IDH2 cells and 10T parental cells treated with ER stress inducer tunicamycin were lysed and protein levels of Bip and CHOP were measured by Western blot. (D) Vector, WT or mutant IDH2 cells were lysed under normoxia or after 6
reinforce with DNA methylation to modulate gene expression and facilitate the transformation by mutant IDH (Figure 4.6D).
Figure 4.6: IDH mutation induces DNA hypermethylation.
(A) and (B) Stacking barplots showing percentage of hyper and hypomethylated differentially methylated cytosines out of all covered CpGs for each chromosome comparing IDH mutant vs. WT and IDH WT vs. vector. Green represents proportion of hypomethylated cytosines and magenta represents hypermethylated ones. Only CpGs with q-value<0.01 and methylation difference of at least 25% are shown. (C) Pathway enrichment analysis of hypermethylated genes in mutant IDH2 cells. (D) Histones were acid-extracted from vector, WT or mutant IDH2 cells at sparse. Levels of histone methylation were measured by Western blot.
Discussion
The research on IDH mutations has been hampered by a lack of robust model system. We previously found that IDH mutation blocked adipocyte differentiation from mouse
fibroblasts (Lu et al., 2012). This finding has been extended to the hematopoietic system with the observations that mutant IDH and 2HG could impair EPO-induced erythrocyte differentiation in an erythroleukemic cell line (Losman et al., 2013) and that
hematopoietic-specific conditional knock-in IDH1 R132H mutant mice had an expansion in early haematopoietic progenitor/stem cell population (Sasaki et al., 2012b).
Nevertheless, none of the previous studies reported in vivo tumorigenecity of mutant IDH, which led to the hypothesis that IDH mutation is insufficient for tumorigenesis. We
report here that using a non-transformed murine mesenchymal multipotent cell line, expression of an IDH mutant not only arrested cells from differentiating into adipocytic and chondrocytic lineages, but also resulted in loss of contact inhibition and tumor formation in vivo. It should be noted that while differentiation impairment was observed shortly after IDH2 mutant expression, contact inhibition required >15 passages of cells in culture, suggesting that the later phenotype may be the result of gradual epigenetic silencing of key genes involved in cell-cell contact.
The common feature shared by all IDH mutations is the production of 2HG. Although the truncated reverse reaction by mutant IDH also consumes NADPH, its impact on total cellular NADPH pool is likely minimal due to a much slower rate compared to the forward reaction by WT IDH(Dang et al., 2009). Importantly, exogenous 2HG could recapitulate mutant IDH’s effect on blocking cell differentiation (Losman et al., 2013; Lu et al., 2012). These data have led to the belief that 2HG is the major mediator of mutant
IDH’s oncogenic potential. Despite a strong link between 2HG and αKG-dependent chromatin modifying enzymes, 2HG is a competitive inhibitor for several additional
αKG-dependent enzymes in vitro (Xu et al., 2011) and in certain context, its impact on
enzymes such as prolyl hydroxylases could lead to important biological consequences (Koivunen et al., 2012; Sasaki et al., 2012a). We therefore tested mutant IDH’s effect on collagen processing and HIF signaling in our system. The data suggest that HIF signaling was unaffected. Although a mild defect in collagen processing was observed, it did not significantly contribute to the loss of contact inhibition by mutant IDH2. In contrast, DNA methylation profiling at base resolution identified a profound chromosome-wide increase in hypermethylated cytosines in IDH2 mutant cells, presumably due to the inhibition of TET family enzymes. There was also global elevation in histone methylation most likely resulting from 2HG’s inhibition of Jumonji-C histone
demethylases. Importantly, gene enrichment analysis identified groups of genes involved in contact inhibition that were over-represented in hypermethylated loci. Collectively our findings suggest that chronic inhibition of TET and Jumonji-C histone demethylase may lead to DNA hypermethylation and permanent silencing of genes that are required for proper response to contact inhibition.
There is a strong interest in developing inhibitors that specifically abrogate mutant IDH’s ability to produce 2HG. It was shown that 2HG’s effect on differentiation arrest could be reversed with delayed kinetics (Losman et al., 2013), suggesting that the epigenetic abnormality is potentially correctable. It remains to be tested, using our model system, whether 2HG is required for tumor initiation and/or maintenance in vivo. Given 2HG’s
remarkable impact on epigenome, it would also be interesting to test whether compounds targeting chromatin modifiers are selectively toxic to tumors with IDH mutation.